TECHNICAL NOTE

HPLC method for rapidly following biodieselfuel transesterification reaction progressusing a core-shell column

Samuel J. Allen & Lisa S. Ott

Received: 20 February 2012 /Revised: 26 April 2012 /Accepted: 2 May 2012 /Published online: 20 May 2012# Springer-Verlag 2012

Abstract There are a wide and growing variety of feedstocksfor biodiesel fuel. Most commonly, these feedstocks containtriglycerides which are transesterified into the fatty acid alkylesters (FAAEs) which comprise biodiesel fuel. While thetranesterification reaction itself is simple, monitoring the re-action progress and reaction products is not. Gas chromatog-raphy–mass spectrometry is useful for assessing the FAAEproducts, but does not directly address either the tri-, di-, ormonoglycerides present from incomplete transesterification orthe free fatty acids which may also be present. Analysis of thebiodiesel reaction mixture is complicated by the solubility andphysical property differences among the components of thetranesterification reaction mixture. In this contribution, wepresent a simple, rapid HPLC method which allows for mon-itoring all of the main components in a biodiesel fuel trans-esterification reaction, with specific emphasis on the ability tomonitor the reaction as a function of time. The utilization of arelatively new, core-shell stationary phase for the HPLC col-umn allows for efficient separation of peaks with short elutiontimes, saving both time and solvent.

Keywords Biodiesel fuel . HPLC . Analysis of reactionintermediates . Core-shell column

Introduction

For both environmental and energy security reasons, bio-diesel fuel is growing in popularity. Biodiesel fuel is

composed of fatty acid alkyl esters, usually prepared bythe alkali-catalyzed transesterification of triglycerides withprimary alcohols at elevated temperature (though other cat-alysts [1, 2] and alcohols are also possibilities). Biodieselfuel is now commercially available in 47 states [3]. Current-ly, commercialized biodiesel fuel relies on food-based feed-stocks (i.e., soy and rapeseed (canola) oil). Since makingbiodiesel from food-based feedstocks both diverts food fromthe table and raises food prices, significant scientific effort isbeing focused on finding non-food sources for biodieselproduction. To that end, a variety of sources have beenexplored, including algae [4], jatropha [5], and other“novelty” sources related to food such as spent coffeegrounds [6] and butter [7].

No matter the feedstock used, the fuels must meetminimum standards for composition and purity beforebeing marketed. One of the primary concerns is theamount of free and bound glycerol in the finished fuel.Free glycerol may be present from incompletely wash-ing the finished fuel; bound glycerides are the mono-and diglycerides present from incompletely transesterify-ing the triglyceride feedstock. Frequently, the presenceof mono- and diglycerides can be circumvented byallowing longer reaction times for transesterification.However, there are economic and life cycle assessmentenergy concerns [8] in maintaining the transesterificationreaction at elevated temperatures for long times. Hence,rapid evaluation of the transesterification reaction prog-ress is essential for further development of biodiesel asa feasible alternative energy source.

For evaluation of transesterification reaction progress,both high temperature gas chromatography (GC) andhigh performance liquid chromatography (HPLC) havebeen employed. HPLC is often considered to be supe-rior to GC for this purpose because no derivatization

S. J. Allen : L. S. Ott (*)California State University, Chico,400 W First Street,Chico, CA 95929, USAe-mail: lsott@csuchico.edu

Anal Bioanal Chem (2012) 404:267–272DOI 10.1007/s00216-012-6094-4

(i.e., trimethyl silylation or acetylation of the free hy-droxy groups) is required—a time and labor intensivestep. However, HPLC analysis of biodiesel transesterifi-cation reaction mixtures is not simple due to the numberof different chemical species present and their physicalproperty differences (including volatility and solubility).

There are several reports in the literature detailingthe difficulty of analyzing transesterification reactionmixtures, which can include glycerol, free fatty acids,fatty acid alkyl esters, and mono-, di- and triglycerides.To add additional complexity, each of the long-chainspecies can have a number of different chain lengthsand/or degrees of unsaturation. In order to detect andquantify all of these species, a careful study by Csern-ica and Hsu examined biodiesel transesterification re-action mixtures using reversed-phase HPLC withrefractive index detection (RID) [9]. RID offers theadvantage of detecting all reaction components, not justthose that have an absorbance in the UV region (themost common mode of detection for biodiesel reactionmixtures). While they were able to detect all the com-ponents in the mixture and quantify many of them,three different solvent programs were used in order tokeep elution times reasonable (60 min or less) whilepreventing overlapping peaks. This approach, then, isnot suitable for a rapid one-pot analysis of the entirereaction mixture.

A one-pot analysis of transesterification reaction mix-tures was successfully carried out using normal phaseHPLC setup, using both a Hypersil column [10] and acyanopropyl column [11]. However, the method for thisanalysis uses methyl tert-butyl ether (MTBE) as a sol-vent, which can have significant environmental concernif not disposed of properly. Additionally, this methodemploys an evaporative light scattering detector (ELSD).While the ELSD does offer the advantage of being responsiveto all chemical species, this detector is less commonly foundin research laboratories.

Finally, there are some reports detailing the use ofreversed-phase HPLC with UV detection: we foundthese reports to be the most intriguing since this is themost common, least expensive instrumental setup forHPLC. A statistically optimized method using a C18

column, a linear ternary gradient (using water, acetoni-trile, and hexane/isopropanol solvents) and UV detec-tion at 210 nm was reported [12] and expanded upon[13]. This method reports successful separation of thetranesterification reaction components in approximately30 min. We were unable to replicate the 30-min elu-tion time, however. We did achieve complete elutionwith sufficient resolution using a slightly different sol-vent program [14]; however, this method took more than45 min for complete elution on a traditional C18 column

(see “Experimental” section for details on this solventprogram).

These longer elution times prompted our interest inusing similar methods with new, core-shell type HPLCpacking materials. These column materials are designedto increase column efficiency with shorter run timescompared to a traditional C18 porous particle columns[15–17]. Consequently, we examined the performance ofcolumn packed with 2.6 μm core-shell particles. Theseparticles have a 1.9 μm nonporous silica core and a0.35 μm porous silica shell. The silica shell is function-alized with C18 chains and coated with butyl side chainsto enhance peak shape. These columns are purported tooffer UHPLC-like efficiency on a standard HPLC sys-tem. Overall, we find that the core-shell column exam-ined herein provides better peak separation, narrowerpeak width, and shorter run times for analysis of thetransesterification reaction mixture compared with tradi-tional C18 columns.

450

950

1450

1950

2450

0 5 10 15 20 25 30

Inte

nsity

(AU

)

Time (min)

(a)

FFAs

Methyl Esters

-50

150

350

550

750

950

1150

1350

1550

0 1 0 2 0 3 0 4 0 5 0

AU

Time (min)

Diglycerides

Triglycerides

(b)

Fig. 1 Chromatograms showing (a) almond oil-derived biodiesel and (b)almond oil, both acquired using the C18 column. The two strong peaks formethyl esters are the methyl oleate (17.2 min) and methyl linoleate(19.0 min); the methyl palmitate is visible as a small peak at 15.5 min

268 S.J. Allen, L.S. Ott

Experimental

The biodiesel fuel used in this work to study the transester-ification reaction mixture was prepared from almonds grownon California State University Chico’s University Farm (fur-ther details on almond oil-derived biodiesel will be reportedelsewhere [18]). The almonds were dried in a 110 °C ovenovernight, then finely ground with a commercial coffeegrinder. The oil was isolated via Soxhlet extraction withn-hexane for 24 h1. After removal of the n-hexane solventwith rotary evaporation, portions of the oil were transesterifiedto biodiesel fuel. The transesterification reaction conditionsincluded a 4:1 vol/vol ratio of oil to methanol (Fisher Scien-tific, ≥99.9 % purity), an approximately 3 % by weight sodi-um hydroxide catalyst (Fisher Scientific, >95 % purity), and atemperature of 55 °C. Using methanol as the alcohol in thetranesterification reaction results in the formation of fatty acidmethyl ester (FAME) products.

As part of this study, aliquots of the reaction solutionwere removed in order to monitor reaction progress versustime by HPLC. At a pre-determined time, an approximately20-μL aliquot of the reaction solution was removed andquenched by addition to 4 mL cold hexanes to yield a1:200 dilution. To confirm that the quenching proceduresufficiently halted the transesterification reaction progress,1H NMR spectroscopy was used to monitor the methyl esterpeak of the FAME product relative to an internal benzenestandard over a 24-h period. For all samples studied, therewas no change in the spectra over a 24-h period. Addition-ally, analysis of selected mixtures up to 1 week later con-firmed that no changes in the spectra took place.

All solvents used for chromatography were HPLC-grade,vacuum-suctioned through a 0.45 μm filter prior to use, andstored in a refrigerator when not in use. An Agilent 1100HPLC system with UV detection at 210 nm was employed.Two columns were tested: a “traditional” porous particle C18

column, and the relatively new core-shell non porous particlecolumn discussed earlier. The traditional column was a Dio-nex C18, 250×4.6 mm, with a 5 μm particle size. The core-shell column used was a Kinetex XB-C18, 100×4.6 mm, witha 2.6 μm particle size (Phenomenex, Torrance CA). Bothanalytical columns were protected by guard columns.

For analysis, 1:40 vol/vol aliquots of the transester-ification reaction mixture were dissolved in 4:5 hexane/isopropanol. A 20-μL sampling loop was flushed threetimes with the analyte solution before loading a samplefor analysis. A similar procedure was carried out forsolutions of reference compounds.

For both columns, the solvent program was 70 % to 30 %vol/vol ratio of acetonitrile/water at 0 min, linearly decreas-ing the amount of water to 100 % acetonitrile at 10 min,holding at 100 % acetonitrile until 20 min, and finally aballistic increase to 50 % acetonitrile, 50 % 4:5 hexane/isopropanol from 20 to 25 min. For the traditional C18

column, this solvent program was maintained until thetriglyceride peaks had eluted. Typical flow rates were1.3 mL/min for the traditional column and 1.0 mL/min forthe core-shell column.

Standards used for retention time determination oftriglycerides, diglycerides, monoglycerides, fatty acidmethyl esters, and free fatty acids were purchased fromSigma-Aldrich in stated grades of 99 % purity orgreater and used as received. These standards includedstearic acid, oleic acid, linoleic acid, methyl stearate,methyl oleate, methyl linolenate, monoolein, diolein,and triolein. Many of the standards used were selectedfrom the “oleate” family (commonly denoted in theliterature as “18:1”, with the first number referring tothe chain length and the second referring to the numberof sites of unsaturation) since the almond oil employedin this study is rich in the 18:1 species. Additionally,the 18:1 species is common component isolated frommany biodiesel feedstocks, and is considered a desir-able component in terms of finished fuel properties2.

1 Caution should be exercised when using n-hexane in the laboratory;this solvent can have a negative effect on the peripheral nervous system(and, with continued exposure, the central nervous system). Usersshould consult a materials safety data sheet before use of n-hexane.

2 For GC-MS analysis, an approximately 10 μL aliquot of the samplewas diluted in approximately 1 mL n-hexane solvent. The column usedfor analysis was 30 m in length with 0.250 mm ID and a 0.15 μLstationary phase composed of (50 % cyanopropyl)-methylpolysilox-ane. The method parameters included a 1-μL injection volume, 50:1split, inlet temperature of 325 °C, and mass spectrometric detectionscanning from 15–550 amu. The oven parameters were as follows:initial 80 °C for 2.00 min, ramp 15 °C/min to 190 °C, ramp 3.5 °C/minto 240 °C.

0

50

100

150

200

250

300

350

400

450

0 5 10 15 20 25

Inte

nsit

y (A

U)

Retention time (min)

FFAs

Methyl Esters

Diglycerides

Triglycerides

Fig. 2 Chromatogram of an incompletely transesterified biodiesel fuelformation reaction. The small peaks between 0–6 minutes are a mixtureof FFAs and monoglycerides. The two main FAME peaks observed arefor methyl oleate (10.1 min) and methyl linoleate (11.8 min)

Rapid HPLC Monitoring of biodiesel fuel preparation 269

Results and discussion

For either a traditional column or the core-shell column, theretention time of individual species will vary based on thechain length and degree of unsaturation of the chains [9].However, many of the most common (and by some criteria,the most desirable [19]) components of biodiesel fuel havesimilar composition profiles. For example, the almond oilbiodiesel used as a test case in the current work has threemain components, as determined in a separate experimentby GC with mass spectrometric detection2: methyl palmitate(16:0; 6.5 % of total), methyl oleate (18:1; 66.2 % of total),

and methyl linoleate (18:2; 25.6 % of total). The elutiontimes of these closely related compounds will fall within awindow (2–4 min) of retention times. A retention timewindow is sufficient for evaluating reaction progress (wherewe are mainly trying to identify the presence or absence of acertain class of molecules). At 210 nm, the majority of theabsorbance for biodiesel species comes from the doublebonds on the chains. Species like methyl palmitate, whichdo not have a double bond, are not projected to have astrong absorbance. However, we do observe weak peaksfor these species which are likely attributable to absorbanceby the ester [20].

Fig. 3 HPLC chromatogramsof the biodiesel reactionmixture as a function of time.Little change in thechromatogram is observed afterthe 10 min reaction time aliquot

270 S.J. Allen, L.S. Ott

Using a traditional C18 column and the solvent programdescribed above, adequate separation between free fatty acid(FFA), FAME, and different glyceride species is observed.However, the di- and triglyceride species are stronglyretained on this column, and typical run times for completeelution were approximately 45 minutes (see Fig. 1).

Using the same solvent program, the core-shell columnproduced excellent separation of FFAs, FAMEs, diglycer-ides, and triglycerides with a total elution time of only24 min (just over half of the time required for the traditionalcolumn, see Fig. 2). Suites of peaks within the retentionwindow indicate the presence of the main components in thealmond oil-derived biodiesel fuel. Supplementary experi-ments show that free glycerol elutes at approximately0.8 min using the core-shell column and solvent programdescribed herein. Since glycerol absorbs weakly at the210 nm wavelength used [20], some users may wish touse another method to quantify small amounts of glycerolpresent in biodiesel transesterification reaction mixtures.

This rapid elution allows for complete HPLC monitoringthe progress of a transesterification reaction within a typical 8-h workday (with manual sample introduction; automated sys-tems would easily complete this sequence overnight). Figure 3shows ten chromatograms, each one an aliquot of a trans-esterification reaction mixture. Using a core-shell column,these chromatographic runs can be completed in approximate-ly 4 h. This method allows the user to watch the triglyceridepeaks of the starting material disappear, the diglycerides growin and then fade out, and the appearance of the FAME peaksand minor FFA peaks. Using a traditional column, this anal-ysis would have taken approximately 7 h and, with the longerrun times and slightly higher flow rate, would have used 2.3times more solvent.

Frequently, shorter run times correspond to lower resolution.The core-shell columnwith a shorter run time used in the presentstudy offers comparable resolution to the C18 column with alonger run for species like the FAMEs shown herein (R03.7 forthe core-shell column,R03.6 for the C18 column). However, theresolution of isomers of multiple bonds does suffer on the core-shell column. This is a disadvantage if the goal of the analysisis to quantify each triglyceride species. However, if the maingoal of the analysis is to determine the presence or absenceof triglyceride peaks, this loss of resolution is not a primaryconcern. In addition, the core-shell column offers the benefitof much narrower peak widths. For just one example, thepeaks for the triglycerides in oil samples are two- to threefoldwider on the C18 column than the core-shell column.

Conclusions

The core-shell column, used for monitoring the biodieseltransesterification reaction progress, results in good separation

of the main components of the reaction mixture and shortelution times. This method uses an unmodified HPLC systemwith the most common detector present on HPLC systems,making the method accessible to most research and teachinglabs. The core-shell column is a simple substitution with acomparable cost to a traditional column. Utilization of a core-shell column saves both time and reagents.

Acknowledgments The authors would like to gratefully recognizefinancial support for this work from a California State UniversityProgram for Education and Research in Biotechnology (CSUPERB)Faculty Seed Grant and from California State University, Chico’sInternal Research Grant program.

References

1. Ghaly AE, Dave D, Brooks MS, Budge S (2010) Production ofbiodiesel by enzymatic transesterification: review. Am J BiochemBiotech 6:54–76

2. Cheng ZZ, Lin SS, Lei R, Yan XC, Nie YJ (2011) Synthesis ofbiodiesel from used cooking oils catalyzed by solid acid. Adv MatRes 236–238:496–500

3. http://www.biodiesel.org/buyingbiodiesel/retailfuelingsites/(accessed January 9, 2012)

4. Knothe G (2011) A technical evaulation of biodiesel from vegeta-ble oils vs. algae. Will algae-derived biodiesel perform? GreenChem 13:3048–3065

5. Koh MY, Ghazi TIM (2011) A review of biodiesel production fromJatropha curcas L. oil. Renew Sust Energ Rev 15:2240–2251

6. Kondamudi N, Mohapatra SK, Misra M (2008) Spent coffeegrounds as a versatile source of green energy. J Agric Food Chem56:11757–11760

7. Haas MJ, Adawi N, Berry WW, Feldman E, Kasprzyk S, RatiganB, Scott K, Bockian Landsburg E (2010) Butter as a feedstock forbiodiesel production. J Agric Food Chem 58:7680–7684

8. Cherubini F, Bird N, Cowie A, Jungmeier G (2009) Energy andgreenhouse gas-based LCA of biofuel and bioenergy systems: keyissues, ranges, and recommendations. Resour Conserv Recycl53:434–447

9. Csernica SN, Hsu JT (2010) Simple and efficient method for theanalysis of transesterification reaction mixtures for biodiesel pro-duction by reversed-phase high performance liquid chromatogra-phy. Energy Fuel 24:6131–6141

10. Foglia TA, Jones KC, Phillips JG (2005) Determination of biodieseland triacylglycerols in diesel fuel by LC. Chromatographia 62:115–119

11. Foglia TA, Jones KC, Nuñez A, Phillips JG, Mittelbach M (2004)Comparison of chromatographic methods for the determination ofbound glycerol in biodiesel. Chromatographia 60:305–311

12. Di Nicola G, Pacetti M, Polonara F, Santori G, Stryjek R (2008)Development and optimization of a method for analyzing biodieselmixtures with non-aqueous reversed phase liquid chromatography.J Chromatog A 1190:120–126

13. Santori G, Arteconi A, Di Nicola G, Moglie M, Stryjek R (2009)Quantitation of compounds in biodiesel mixtures with reversed-phase liquid chromatography. Energy Fuel 23:3783–3789

14. Holčapek M, Jandera P, Zderadička P, Hrubá L (2003) Characteriza-tion of triacylglycerol and diacylglycerol composition of plant oilsusing high performance liquid chromatography-atmospheric pressurechemical ionization mass spectrometry. J Chromatog A 1010:195–215

Rapid HPLC Monitoring of biodiesel fuel preparation 271

15. Gritti F, Leonardis I, Shock D, Stevenson P, Shalliker A, GuiochonG (2010) Performance of columns packed with the new shellparticles, Kinetex-C18. J Chromatog A 1217:1589–1603

16. Gritti F, Guiochon G (2010) Performance of columns packed withthe new shell Kinetex-C18 particles in gradient elution chromatog-raphy. J Chromatog A 1217:1604–1615

17. Oláh E, Fekete S, Fekete J, Ganzler K (2010) Comparative studyof new shell-type, sub 2 μm fully porous and monolith stationary

phases, focusing on mass-transfer resistance. J Chromatog A1217:3642–3653

18. Allen SJ, Shelton TS, Ott, LS. in preparation19. Knothe G (2008) “Designer” biodiesel: optimizing fatty ester

composition to improve fuel properties. Energy Fuel 22:1358–1364

20. Bruno TJ, Svoronos PDN (2006) CRC handbook of fundamentalspectroscopic correlation charts. CRC, Boca Raton, FL

272 S.J. Allen, L.S. Ott